By combining high-throughput experimentation and machine learning technologies, the researchers developed the first iridium-catalyzed cross-dimerization of two sulfoxonium ylides, leading to various unsymmetrical E-alkenes. Good yield and excellent chemoselectivity and functional group compatibility under mild conditions enable this protocol to be a promising approach for the synthesis of valuable amide-, ketone-, ester-, and N-heterocycle-substituted unsymmetrical E-alkenes. Amide-sulfoxonium ylide and the Iridium catalyst are both the essential carbene precursors.

The Ir-catalyst can “distinguish” two different sulfoxonium ylides based on their differences in intrinsic reactivity and electronic effect, resulting in highly selective cross-coupling and the formation of high E-selectivity olefin products.

More importantly, they have built a ML-based model (XGB-MAF) that can accurately predict reaction yields over a range of diverse unseen substrates, and also successfully explored the reaction space (up to 600 reactions). Thus, they expect that this HTE-ML workflow will facilitate the development and application of newly metal-catalyzed cross-coupling reaction and sulfoxonium ylide chemistry, which is likely to be of broad interest to organic and medicinal chemistry researchers in both academic and industrial settings.

The Development of ML-Based Models for yield prediction

Image Source：Angew. Chem., DOI: 10.1002/anie.202313638